In the video https://www.youtube.com/watch?v=7Ys_yKGNFRQ, CalTech has a camera capable of filming the picoseconds it takes for light to travel 1mm in distance.

It can do 10 Trillion frames per second.

It cannot be a rotating camera as the fastest they can do is a billion frames per second or so (used to capture nuclear detonations).

How can the film at these speeds?

  • $\begingroup$ The generic principle seems to be this. You are interested in acquiring 3D data labeled by coordinates (x,y,t). You obtain two projections of the information (after using beamsplitter to make two copies of the original light collected) onto 2D, one with coordinates (x,y) (same as what you will see with a normal camera after exposure) and the other with coordinates (f(x),g(x)+t). And somehow using some clever transformation, you can recover the full 3D data (x,y,t) from the two sets of projected 2D data. The transformation, and what f(x) and g(x) are, seem to me a little hard to understand. $\endgroup$ – wcc Jun 4 at 23:30
  • $\begingroup$ It seems the concept of "compressive sensing" is crucial to the technique, and this concept is also central to the demonstration of single-pixel camera. Compressive sensing is a way to go beyond the Nyquist sampling theorem with small number of sampling, by using information of sparsity of data representation in some basis. I am still trying to understand how image reconstruction by compressive sensing works . But assuming you understand this part, you can understand ultrafast frame rate simply as coming from mapping time coordinate into a spatial coordinate using a streak camera. $\endgroup$ – wcc Jun 7 at 18:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.